Insulin Suppresses Transactivation by CAAT/Enhancer-binding Proteins β (C/EBPβ)

CAAT/enhancer-binding proteins (C/EBPs) play an important role in the regulation of gene expression in insulin-responsive tissues. We have found that a complex containing C/EBPβ interacts with an insulin response sequence in the insulin-like growth factor-binding protein-1 (IGFBP-1) gene and that a C/EBP-binding site can mediate effects of insulin on promoter activity. Here, we examined mechanisms mediating this effect of insulin. The ability of insulin to suppress promoter activity via a C/EBP-binding site is blocked by LY294002, a phosphatidylinositol 3-kinase inhibitor, but not by rapamycin, which blocks activation of p70S6 kinase. Dominant negative phosphatidylinositol 3-kinase and protein kinase B (PKB) block the effect of insulin, while activated PKB suppresses promoter function via a C/EBP-binding site, mimicking the effect of insulin. Coexpression studies indicate that insulin and PKB suppress transactivation by C/EBPβ, but not C/EBPα, and that N-terminal transactivation domains in C/EBPβ are required. Studies with Gal4 fusion proteins reveal that insulin and PKB suppress transactivation by the major activation domain in C/EBPβ (AD II), located between amino acids 31 and 83. Studies with E1A protein indicate that interaction with p300/CBP is required for transactivation by AD II and the effect of insulin and PKB. Based on a consensus sequence, we identified a PKB phosphorylation site (Ser1834) within the region of p300/CBP known to bind C/EBPβ. Mammalian two-hybrid studies indicate that insulin and PKB disrupt interactions between this region of p300 and AD II and that Ser1834 is critical for this effect. Signaling by PKB and phosphorylation of Ser1834 may play an important role in modulating interactions between p300/CBP and transcription factors and mediate effects of insulin and related growth factors on gene expression.

CAAT/enhancer-binding proteins (C/EBPs) 1 play an important role in the regulation of gene expression in insulin-responsive tissues (1,2). Studies in cell culture and with knockout mice indicate that C/EBP␣ and C/EBP␤ both contribute to the regulation of hepatic glucose production (3)(4)(5), including the multihormonal regulation of phosphoenolpyruvate carboxykinase (PEPCK) (6 -10), a rate-limiting enzyme for gluconeogenesis. We have reported that a nucleoprotein complex containing C/EBP␤ interacts with oligonucleotide probes containing insulin response sequences (IRSs) from the insulin-like growth factor-binding protein-1 (IGFBP-1) and PEPCK genes (11). This complex interacts with the IRS in a sequence-specific fashion that correlates with the ability of the IRS to mediate effects of insulin on promoter activity, and replacing the IRS with a consensus C/EBP-binding site in reporter gene constructs maintains the effect of insulin (11). Together, these results suggest that signaling to C/EBP proteins may contribute to the effects of insulin on gene expression via an IRS. Based on these observations, we examined specific mechanisms by which insulin may suppress promoter activity via C/EBP proteins.
Previous studies have shown that insulin inhibits basal activity of the IGFBP-1 promoter via an IRS, and that this effect of insulin is mediated by protein kinase B (PKB) downstream from phosphatidylinositol 3Ј-kinase (PI 3-kinase) (12). Activated PKB is translocated to the nucleus where it may interact with nuclear targets directly involved in the regulation of gene expression (13). PKB has been shown to phosphorylate and suppress transactivation by a subset of forkhead transcription factors, including FKHR, FKHRL1, and AFX (14 -16). Since a nucleoprotein complex containing C/EBP␤ interacts with IRSs in the IGFBP-1 and PEPCK genes, we considered the possibility that C/EBP proteins also might contribute to the ability of insulin to regulate promoter activity downstream from PI 3-kinase and PKB.

EXPERIMENTAL PROCEDURES
Plasmid Constructs-The SauI/HgaI fragment of the IGFBP-1 promoter which extends 320 base pairs 5Ј from the RNA cap site was cloned into pGL2 (Promega) (BP1.Luc) and modified to create the ⌬IRS.1, ⌬IRS.1M, and ⌬C/EBP reporter gene constructs, as previously reported (11,12). The pAlb(DEI) 4 reporter gene construct (32) contains 4 copies of a naturally occurring C/EBP-binding site placed immediately upstream from the albumin minimal promoter, and the pG5e1b construct consists of 5 Gal4-binding sites introduced immediately upstream of the e1b minimal promoter. Expression vectors for C/EBP proteins and fusion proteins containing the Gal4 DNA-binding domain have been described (32). Expression vectors for 12S E1A and ⌬E1A, where amino acids 2-36 are deleted, were provided by Dr. Pradip Raychaudhuri (33). The ⌬p85 expression vector was provided by Dr. M. Kasuga (34). Vectors expressing kinase-defective (Lys 179 -PKB) and constitutively active, myristoylated PKB (Myr-PKB) were provided by Dr. Thomas Franke (35). Mammalian expression vectors for CHOP and CHOP-LZ Ϫ were provided by Dr. David Ron (36).
Site-directed Mutagenesis of p300 and Creation of Fusion Proteins-The full-length p300 cDNA was provided by Dr. David Livingston (37). The region coding for amino acids 1752-1859 region was amplified by polymerase chain reaction with one primer which contains an XhoI site (5Ј-CAGCTCGAGACCATGGGGATCCTCAATTGCTCACTGCCATCC-TGC-3Ј) and a second primer (5Ј-GGCTCTAGACTAAGTTGGTGTCG-TTGGAGTGGCAGGAG-3Ј) which contains a stop codon and an XbaI site. This fragment was cloned into the XhoI/XbaI site in pAlter.Max vector (Promega) and single stranded DNA was prepared with helper phage for site-directed mutagenesis. A BamHI site was introduced at the 5Ј end of this fragment and Ser 1834 was replaced with alanine or aspartate by site-directed mutagenesis using the following oligonucleotides: p300.BamHI, 5Ј-P-CAGCGGCGGCGGGGATCCCCGGCGGGTG-CTGC-3Ј; S1834A, 5Ј-P-CACACCAGTCCGCTGCATGGCGGCCATCC-TCCTGCGAAG-3Ј; S1834D, 5Ј-P-CACACCAGTCCGCTGCATGTCGG-CCATCCTCCTGCGAAG-3Ј. The sequences of the polymerase chain reaction product and its mutations were confirmed by DNA sequencing in the University of Illinois at Chicago DNA Sequencing Center.
BamHI-XbaI fragments coding for amino acids 1752-1859 of p300 with/without mutation of Ser 1834 were subcloned in-frame with the VP16 activation domain in the pVP16 eukaryotic expression vector (CLONTECH). A copy of the SV40 nuclear localization signal has been introduced in-frame with the VP16 activation domain in this vector by the manufacturer to ensure that expressed proteins are transported to the nucleus. Using a NotI site in the pAlter polylinker, BamHI-NotI fragments also were excised and cloned in-frame with glutathione Stransferase (GST) in pGEX4T-3 (Amersham Pharmacia Biotech). pGEX4T-3 vectors were introduced into pBL21RIL bacteria (Stratagene) and fusion proteins were induced at 30°C with isopropyl-1-thio-␤-D-galactopyranoside. Cells were harvested by centrifugation and lysed by sonication, and lysates were cleared by centrifugation.
In Vitro Kinase Studies-For kinase studies with synthetic peptides, a peptide containing residues 1826 -1842 of p300 (QMLRRR-MASMQRTGVVG) was synthesized, purified by high performance liquid chromatography, and analyzed by mass spectroscopy at the UIC Protein Synthesis Center. A peptide (RPRAATF) derived from a known PKB phosphorylation site in glycogen-synthase kinase-3 was purchased from Upstate Biotechnology, Inc. HepG2 cells were transfected with expression vectors for HA-tagged Myr-PKB and Myr-Lys 179 -PKB, and tagged proteins were immunoprecipitated with monoclonal anti-HA antibodies (Calbiochem) and protein G-Sepharose (Amersham Pharmacia Biotech), as before (12). Beads were washed and then incubated with target peptides for 10 min at 30°C with [␥-32 P]ATP (Amersham Pharmacia Biotech) using buffers and peptide inhibitors of PKA and calciumdependent kinase provided with the Akt kinase kit (Upstate Biotechnology, Inc.). Kinase reactions were terminated by heating at 100°C for 5 min, and phosphorylation was measured by scintillation counting after spotting onto p81 phosphocellulose discs and washing with phosphoric acid, as before (12).
For phosphorylation studies with GST-p300 fusion proteins, bacterial lysates containing ϳ3 g of fusion protein were incubated with 25 l of a 50% slurry of washed glutathione-agarose beads (Amersham Pharmacia Biotech). Washed beads were equilibrated with 50 l of kinase buffer containing 0.25 units of active PKB␣, provided by Dr. P. Cohen. Phosphorylation reactions were initiated at 30°C by the addition of [␥-32 P]ATP and terminated 20 min later by addition of 5 ϫ Laemmli sample buffer and heating at 100°C. Samples were cleared by centrifugation and loaded for 4 -20% gradient SDS-polyacrylamide gel electrophoresis. Gels were stained with Coomassie Blue, then dried and phosphoproteins identified by autoradiography at Ϫ70°C with enhancing screens.
Cell Culture, Transfection, and Reporter Gene Analysis-HepG2 cells in 60-mm dishes were transfected in triplicate with calcium phosphate precipitates containing reporter gene and expression vectors together with appropriate amounts of empty vector, as previously reported (12,14). Cells were refed with Dulbecco's modified Eagle's medium plus 1 g/liter fatty acid-free bovine serum albumin (Sigma) with/without 100 nM recombinant human insulin (Sigma) and/or 50 M LY294002 (Calbiochem) or 200 nM rapamycin (Sigma) 18 h prior to the preparation of lysates and analysis of luciferase activity (12).

RESULTS
Insulin inhibits promoter activity in luciferase reporter gene constructs containing an IRS (CAAAACA; ⌬IRS.1) or when the IRS is replaced by a consensus binding site for C/EBP proteins (TTGCGCAA; ⌬C/EBP), but not when the IRS is replaced by an unrelated sequence (⌬IRS.1M) (Fig. 1). This indicates that the ability of insulin to suppress promoter activity is sequencespecific and can be mediated by either an IRS or C/EBP site.
Previous studies have shown that the ability of insulin to inhibit promoter activity via an IRS is mediated through the PI 3-kinase/PKB signaling pathway (12), and we asked whether signaling to C/EBP proteins also might occur through this pathway. As shown in Fig. 1, treatment with LY294002, a specific inhibitor of PI 3-kinase, or coexpression of a dominant negative form of PI 3-kinase (⌬p85) blocks the ability of insulin to suppress promoter activity via either an IRS or C/EBPbinding site (Fig. 1), indicating that this effect of insulin also is PI 3-kinase-dependent. Treatment with rapamycin, which blocks the activation of p70 S6 kinase downstream from PI 3-ki- nase, does not disrupt the ability of insulin to inhibit promoter activity. Coexpression of a kinase-deficient form of PKB (Lys 179 -PKB), which blocks the activation of PKB in HepG2 cells (12), disrupts the ability of insulin to inhibit promoter activity. Expression of constitutively active PKB (Myr-PKB) inhibits promoter activity of reporter gene constructs containing either an IRS (⌬IRS.1) or a C/EBP-binding site (⌬C/EBP), but not an unrelated sequence (⌬IRS.1M), mimicking the effect of insulin. Together, these results indicate that signaling via PKB can mediate effects of insulin on promoter activity via either an IRS or C/EBP-binding site.
To determine whether endogenous C/EBP proteins are required for the ability of insulin to inhibit promoter activity via a C/EBP-binding site, we performed coexpression studies with CHOP. CHOP, a C/EBP family member which contains a leucine zipper dimerization domain but no functional DNAbinding domain, disrupts the ability of endogenous C/EBP proteins to form hetero-or homodimers that are capable of binding to canonical C/EBP sites (11,36). In contrast, CHOP does not prevent C/EBP␤ from interacting with other nuclear proteins and forming a complex with an IRS (11). As shown in Fig. 2, overexpression of CHOP blocks the ability of insulin and Myr-PKB to inhibit promoter activity via a C/EBP-binding site in the ⌬C/EBP construct (solid bars). This effect of CHOP is specific, since CHOP does not disrupt other effects of insulin and Myr-PKB in these cells, including the ability to suppress promoter function via an IRS in the ⌬IRS.1 construct (open bars). CHOP-LZ Ϫ , which is missing the leucine zipper dimerization domain, does not block the ability of insulin and Myr-PKB to suppress promoter function via a C/EBP-binding site, indicating that this effect of CHOP is mediated through interaction with endogenous proteins via its leucine zipper domain. Together, these results indicate that interaction with hetero-or homodimers of C/EBP proteins is necessary for the ability of insulin and PKB to suppress promoter activity via this C/EBPbinding site.
To determine whether insulin and PKB suppress promoter activity when specific C/EBP proteins occupy this C/EBP-binding site, we performed co-transfection studies with expression vectors for C/EBP␣ and C/EBP␤. As shown in the left panel of Fig. 3A, C/EBP␣ and full-length C/EBP␤ (C/EBP␤ 1-276 ) effectively stimulate promoter activity in the ⌬C/EBP construct. This effect is disrupted when the C/EBP site is replaced by an unrelated sequence (⌬IRS.1M), indicating that it is mediated through the C/EBP site. Truncated C/EBP␤ (C/EBP␤ 132-276 ), which is missing two N-terminal transactivation domains, is less effective in stimulating promoter activity and this residual effect is not disrupted when the C/EBP-binding site is replaced by an unrelated sequence (⌬IRS.1M). These results confirm that C/EBP␣ and full-length C/EBP␤ stimulate promoter activity via the C/EBP-binding site and that N-terminal activation domains are required for sequence-specific transactivation by C/EBP␤.
As shown in the right panel of Fig. 3A, coexpression of C/EBP␣ disrupts the ability of insulin and PKB to suppress promoter activity in the ⌬C/EBP construct. In contrast, insulin and PKB suppress the ability of full-length C/EBP␤ (C/EBP␤  to stimulate promoter activity. This effect of insulin and PKB is disrupted when the N-terminal region of C/EBP␤ is deleted (C/EBP␤ 132-276 ). These results indicate that insulin and PKB suppress transactivation by C/EBP␤, but not C/EBP␣, and that the N-terminal region of C/EBP␤ is required for this effect.
We performed similar studies using a reporter gene con- Cells were refed with serum-free medium 18 h prior to harvest. Right panel, HepG2 cells were transfected with 10 g/dish plasmid DNA, including 1 g of the ⌬C/EBP construct together with 5 g of vector expressing C/EBP␣, C/EBP␤ 1-276 , or C/EBP␤ 132-276 with/without 1 g of Myr-PKB expression vector. Cells were refed with serum-free medium with/without insulin 18 h prior to harvest and analysis of luciferase activity. B, pAlb(DEI) 4 reporter gene construct. Cells were transfected with 10 g/dish plasmid DNA, including 1 g of the pAlb(DEI) 4 reporter gene construct with/without 5 g of vector expressing C/EBP␣, C/EBP␤ 1-276 , or C/EBP␤ 132-276 , and/or 1 g of Myr-PKB. Cells were refed with serum-free medium with/without insulin 18 h prior to harvest. struct where an array of 4 naturally occurring C/EBP-binding sites has been placed immediately upstream of the minimal albumin promoter (pAlb[DEI] 4 ). In contrast to the ⌬C/EBP construct, insulin modestly stimulates basal activity of the pAlb(DEI) 4 promoter, possibly reflecting effects which are mediated via other transcription factors that can interact with this naturally occurring C/EBP-binding site or with other parts of this construct. As shown in Fig. 3B, overexpression of C/EBP␣ or C/EBP␤ stimulates the activity of this artificial promoter construct and truncated C/EBP␤ is less effective. Insulin and PKB suppress the ability of C/EBP␤ to stimulate the activity of this artificial construct. In contrast, neither C/EBP␣ nor truncated C/EBP␤ confers this effect of insulin and PKB, confirming that insulin and PKB suppress transactivation by C/EBP␤, but not C/EBP␣, and that the N-terminal region of C/EBP␤ is required for this effect.
We next examined which regions of C/EBP␤ are involved in mediating the effect of insulin and PKB on transactivation. For these experiments, we expressed fusion proteins containing the Gal4 DNA-binding domain (which contains its own nuclear localization signal) in-frame with portions of C/EBP␤, or the activation domain of VP16 as a control. Transactivation was measured using the pG5e1b luciferase reporter gene construct, which contains 5 Gal4-binding sites immediately 5Ј to the e1b minimal promoter and has negligible basal activity in HepG2 cells. Fig. 4A indicates the location of the leucine zipper dimerization domain (LZ), DNA-binding domain, and the two N-terminal transactivation domains previously identified in C/EBP␤ (AD I and II) (32). As shown in Fig. 4B, the Gal 4 fusion protein containing full-length C/EBP␤ stimulates the pG5e1b promoter. Transactivation is enhanced when the C-terminal region containing the leucine zipper and DNA-binding domain is removed from C/EBP␤ (Gal4.␤1-132), and transactivation is increased still further when an additional 39 amino acids are removed (Gal4.␤1-83), consistent with previous studies (32). The N-terminal region of C/EBP␤ contains two activation domains, which are located between amino acids 1-31 (AD I) and 31-83 (AD II). The Gal4.␤31-83 fusion protein is more potent in stimulating promoter activity than the Gal4.␤1-31 protein, consistent with previous studies indicating that the major activation domain in C/EBP␤ is located between amino acids 31 and 83 (32).
As shown in Fig. 4C, insulin and PKB suppress transactivation by Gal4.␤1-276 but not Gal4.VP16, indicating that this effect of insulin and PKB is specific. Insulin and PKB also suppress transactivation by Gal4.␤1-132 and Gal4.␤1-83, indicating that regions outside the N-terminal transactivation domains of C/EBP␤ are not required for this effect of insulin and PKB. Insulin and PKB suppress transactivation by Gal4.␤31-83 (AD II) but do not suppress transactivation by Gal4.␤1-31 (AD I), indicating that this effect is specific for ADII.
Previous studies have shown that p300/CBP coactivator proteins interact directly with the N-terminal region of C/EBP␤ (17). As shown in the left panel of Fig. 5A, coexpression of E1A, an adenovirus protein that binds and sequesters p300/CBP, disrupts transactivation by the Gal4.␤1-132 fusion protein in a dose-dependent fashion. In contrast, coexpression of ⌬E1A, which does not interact with p300/CBP but still binds retinoblastoma protein and related proteins (33,38), does not suppress transactivation by the Gal4.␤1-132 fusion protein but actually enhances this effect. Together, these results indicate that interaction with p300/CBP is critical for transactivation by this region of C/EBP␤, and that other cellular proteins that can interact with ⌬E1A may repress this function, similar to re-sults obtained with other transcription factors (39). 2 As shown in the right panel of Fig. 5A, coexpression of E1A (but not ⌬E1A) also disrupts the ability of insulin and PKB to suppress transactivation by the Gal4.␤1-132 fusion protein.
Together, these results suggest that interaction with p300/CBP is required for effective transactivation by the N-terminal region of C/EBP␤ and for the inhibitory effect of insulin and PKB.
As shown in the left panel of Fig. 5B, E1A (but not ⌬E1A) also suppresses transactivation by AD II of C/EBP␤ (Gal4.␤31-83), but not AD I (Gal4.␤1-31), indicating that interaction with p300/CBP is required for transactivation by AD II, but not AD I. As shown in the right panel of Fig. 5B, E1A (but not ⌬E1A) disrupts the ability of insulin and PKB to suppress transactivation by Gal4.␤31-83, indicating that interaction with p300/ CBP is required for insulin and PKB to suppress transactivation by AD II. Previous studies have shown that C/EBP␤ interacts directly with residues 1752-1859 of p300 and that this interaction involves the N-terminal region of C/EBP␤ that contains ADII (17). As shown in Fig. 6A, this region of p300/CBP, which includes a portion of a cysteine/histidine-rich domain (CH3) and the adjacent glutamine-rich domain (Q), also interacts with a number of other trans-acting factors that contribute to the regulation of gene expression (18 -31).
Based on a consensus sequence for PKB phosphorylation sites (Arg-Xaa-Arg-Xaa-Xaa-Ser/Thr) (40), we identified a putative PKB phosphorylation site in this region of human p300 (Ser 1834 ). This PKB site is conserved in both human and mouse CBP (Fig. 6B), suggesting that it may be physiologically important. The sequence for this region of mouse p300 has not been reported. We did not identify other PKB sites in p300, CBP, or C/EBP␤. Based on this observation, we asked whether this site is phosphorylated by PKB and is required to disrupt interactions between this region of p300 and AD II. As shown in Fig. 7A, activated PKB (but not kinase-defective PKB) phosphorylates a synthetic peptide containing amino acids 1826 through 1842 of human p300 and this phosphorylation is similar to the level achieved with a peptide derived from glycogen-synthase kinase-3, a known substrate for PKB. As shown in Fig. 7B, PKB also phosphorylates a bacterially expressed recombinant fusion protein containing amino acids 1752-1859 of p300 in-frame with GST (GST-p300), but not GST alone. This phosphorylation is blocked when Ser 1834 is replaced by either alanine (GST.p300(S/A)) or aspartate (GST.p300(S/D)), confirming that Ser 1834 is a target for phosphorylation by PKB in vitro.
To examine interactions between this portion of p300 and C/EBP␤, we next performed mammalian two-hybrid studies in HepG2 cells. As shown in Fig. 8A, coexpression of a fusion protein containing amino acids 1752-1859 of p300 in-frame with the VP16 activation domain enhances the ability of the Gal4.␤1-132 fusion protein to stimulate luciferase activity in HepG2 cells transfected with the pG5e1b reporter gene construct in a dose-dependent fashion. No effect is seen when the VP16 activation domain is expressed alone, indicating that this effect is mediated through interaction with this region of p300. The VP16.p300 fusion protein also enhances transactivation by the Gal4.␤32-83 fusion protein (Fig. 8B), but does not enhance transactivation by Gal4.␤1-31 (data not shown), indicating that this region of p300 interacts with AD II, but not AD I in cells.
We next examined whether insulin and PKB disrupt the interaction between this region of p300 and C/EBP␤, and whether the phosphorylation of Ser 1834 is required for this effect of insulin and PKB. As shown in Fig. 8B, coexpression of VP16.p300 enhances transactivation by the Gal4.␤1-132 fusion protein (left panel) and Gal4.␤31-83 (right panel) without disrupting the ability of insulin and PKB to suppress transactivation. This result indicates that signaling via insulin and PKB disrupts interaction between this region and ADII of C/EBP␤ in cells.
Replacing Ser 1834 with aspartate, a negatively charged amino acid (VP16.p300(S/D)) blocks the ability of insulin and PKB to suppress interaction between this region of p300 and  6. Structure of p300/CBP. A, interaction domains in p300/ CBP. The region within p300/CBP that interacts with C/EBP␤ is underlined. Other trans-acting factors that interact with this region and/or other domains within p300/CBP also are shown. B, predicted PKB phosphorylation site. The consensus sequence for PKB phosphorylation sites (RXRXX(S/T)) and its relationship to the sequence flanking Ser 1834 in human p300 is shown. This predicted PKB site is conserved in human and mouse CBP. The sequence for this region of mouse p300 has not been reported (N/A). ADII in cells, but does not impair the ability of the fusion protein with interact with ADII in two-hybrid studies. This result supports the concept that phosphorylation of Ser 1834 is critical for the effect of insulin and PKB, and suggests that this effect of insulin and PKB is not mediated simply by introducing a negative charge at this site. Replacing Ser 1834 with alanine, a neutral amino acid that is not susceptible to phosphorylation, also blocks the ability of insulin and PKB to disrupt interaction between this region of p300 and ADII, supporting the concept that phosphorylation at this site is required for the effect of insulin and PKB. Replacing Ser 1834 with alanine partially reduces the ability of this fusion protein to interact with ADII in the two-hybrid assay (Fig. 8B), supporting the concept that Ser 1834 is important for interaction between this region of p300 and ADII. DISCUSSION C/EBP␤ plays an important role in the regulation of gene expression in the liver, mammary gland, and adipose, hematopoietic and reproductive tissues (41)(42)(43)(44)(45). We recently found that a nucleoprotein complex containing C/EBP␤ interacts with insulin response sequences in the IGFBP-1 and PEPCK genes and that a consensus C/EBP-binding site confers effects of insulin on promoter activity, similar to an IRS (11). Based on these findings, we asked whether insulin may suppress transactivation by specific C/EBP family members and examined mechanisms mediating this effect of insulin.
Studies with pharmacological inhibitors and with constitutively active and dominant negative forms of PI 3-kinase and PKB revealed that PI 3-kinase plays a critical role in mediating effects of insulin on promoter activity via a C/EBP-binding site and that signaling via PKB can mediate this effect of insulin. Previous studies have shown that signaling via PI 3-kinase-dependent pathways mediates effects of insulin on hepatic expression of IGFBP-1, PEPCK, and glucose-6-phosphatase genes (12, 46 -48) and we have reported that insulin suppresses basal IGFBP-1 promoter activity through an IRS via events that are mediated by PKB downstream from PI 3-kinase (12). Recent studies indicate that PKB can mediate effects of insulin on promoter activity via the phosphorylation of FKHR, FKHRL1, or AFX, a subgroup of the forkhead/winged-helix family of transcription factors which contain several PKB phos- HA-tagged, constitutively active, myristoylated PKB (Myr-PKB) and kinase-deficient, myristoylated PKB (Myr-Lys 179 -PKB) were expressed in HepG2 cells, then bound to Sepharose beads by immunoprecipitation with anti-HA monoclonal antibody and incubated with synthetic peptides containing amino acids 1826 -1842 of human p300 (p300) or a peptide containing a known PKB phosphorylation site derived from glycogen-synthase kinase-3 (GSK3). The incorporation of 32 P into phosphopeptide was determined by scintillation counting after spotting on phosphocellulose discs, as described under "Experimental Procedures." B, GST.p300 fusion proteins. GST and fusion proteins containing GST in-frame with amino acids 1752-1859 of p300 (GST.p300) with/without mutation of Ser 1834 to alanine (GST.p300(S/A)) or aspartate (GST.p300(S/D)) were expressed in bacteria, then bound to glutathione-Sepharose beads prior to incubation with/without activated PKB and [␥-32 P]ATP for 20 min at 30°C. Phosphorylation reactions were terminated and glutathione-bound fusion proteins were eluted by the addition of Laemmli sample buffer and heating at 100°C. Proteins were loaded for 4 -20% SDS-polyacrylamide gel electrophoresis, then stained with Coomassie Blue (lower panel) and 32 P-labeled proteins were identified in dried gels by autoradiography (upper panel). phorylation sites (14,15,16,49). The results of the present study indicate that signaling via PI 3-kinase and PKB also can disrupt transactivation by C/EBP␤ and, thereby, suppress promoter activity by a mechanism that is independent of forkhead transcription factors.
This effect of insulin appears to be selective, since insulin and PKB suppressed transactivation by C/EBP␤, but not C/EBP␣. We found that insulin and PKB suppress transactivation by the major activation domain in C/EBP␤, located between residues 31 and 83. It is interesting to note that the corresponding region of C/EBP␣ contains an insertion of 12 amino acids that are not present in AD II of C/EBP␤ (50). It is possible that these additional residues may alter the function of this activation domain, including the ability of insulin and PKB to regulate interactions with coactivator proteins required for transactivation. Also, C/EBP␣ contains another activation domain which is not present in C/EBP␤ (50) and which may not be susceptible to regulation by insulin and PKB. In addition, regions outside the activation domains have been found to contribute to functional differences between C/EBP␣ and C/EBP␤ (32,51) and also might contribute to differences in the effects of insulin and PKB on transactivation by full-length C/EBP␣ and C/EBP␤.
Previous studies have shown that extracellular stimuli can alter the phosphorylation of several sites in C/EBP␤, and modify their effects on promoter activity. Calcium-regulated phosphorylation of Ser 276 in the leucine zipper domain of C/EBP␤ results in enhanced transactivation (52), as does Ras-dependent phosphorylation of Thr 235 , and the phosphorylation of Ser 105 by protein kinase A (53,54) or pp90 Rsk (55). Phosphorylation by protein kinase C of Ser 240 in the DNA-binding domain of C/EBP␤ reduces transactivation by interfering with DNA binding (54). Our studies with Gal4 fusion proteins indicate that phosphorylation at these sites is not required for the ability of insulin and PKB to suppress transactivation by C/EBP␤. Instead, we find that insulin and PKB suppress the function of the major C/EBP␤ activation domain by disrupting interactions with p300/CBP proteins. To our knowledge, this report provides the first evidence for an effect of insulin or PKB that is mediated through the modification of a nuclear coactivator, in this case p300/CBP.
In this context, it is important to note that other kinases also may contribute to this effect of insulin. PKB, like other members of the AGC kinase family, is activated by the phosphorylation of a conserved hydrophobic activation loop by phosphatidylinositoldependent kinase-1 (56,57). We found that overexpression of a kinase-deficient form of PKB, which blocks the activation of PKB in HepG2 cells (12), disrupts the ability of insulin to suppress transactivation by C/EBP␤. It is possible that overexpression of kinase-deficient PKB, which may interfere with the ability of phosphatidylinositol-dependent kinase-1 to phosphorylate and activate PKB, also may block the activation of other AGC kinases downstream from phosphatidylinositoldependent kinase-1. As recently reviewed, several of these kinases can phosphorylate sites that are similar to the consensus motif for PKB, including serum-and glucocorticoidinducible kinase (glycogen-synthase kinase-3), pp90 Rsk , and p70 S6 kinase (56,58). Preliminary studies indicate that pp90 Rsk and p70 S6 kinase also can phosphorylate Ser 1834 in vitro, but not as efficiently as PKB. 3 Additional studies will be required to determine whether other AGC kinases also contribute to signaling to Ser 1834 in vivo.
To explore the possibility that phosphorylation of Ser 1834 is required to disrupt interactions between p300 and C/EBP␤, we performed mammalian two-hybrid studies with VP16.p300 fusion proteins where this residue was altered. Replacing Ser 1834 with either aspartate or alanine disrupts the ability of insulin and PKB to suppress interaction between this domain and ADII in cells, indicating that this site is critical for the effect of insulin and PKB. Interestingly, replacing Ser 1834 with alanine, a neutral amino acid, reduces the ability of VP16.p300 fusion proteins to interact with ADII in two-hybrid studies, while introducing aspartate, a negatively charged amino acid, is neutral in this assay. Preliminary studies indicate that replacing Ser 1834 with alanine also partially reduces the ability of GST fusion proteins containing this region of p300 to bind C/EBP␤ in in vitro pull-down binding assays, while placing an aspartate residue at this location does not alter C/EBP␤ binding either positively or negatively, 3 similar to the results of two-hybrid assays. These results support the concept that Ser 1834 is important for the ability of this region of p300 to interact with C/EBP␤, and indicate that this interaction is not altered simply by introducing a negative charge at this site.
Previous studies have shown that phosphorylation can modify protein function by multiple mechanisms, and the introduction of a negative charge is not always sufficient to mimic the effect of phosphorylation. For example, phosphorylation of Ser 133 in CREB results in the recruitment of p300/CBP, and replacing this residue with aspartate does not reproduce that effect (59). Based on the results of the present study, it is possible that phosphorylation of Ser 1834 may recruit other proteins to this site in p300 and, thereby, block interactions with ADII. Nakajima et al. (23) have reported that insulin treatment causes pp90 Rsk to associate with this region of p300/CBP. Additional studies will be required to determine whether phosphorylation of Ser 1834 is sufficient to alter interactions between p300/CBP and AD II of C/EBP␤, or whether the recruitment of other proteins (including pp90 Rsk ) is required for this effect. p300/CBP can interact with multiple proteins at the same time via distinct domains and, thereby, mediate cooperative effects of transcription factors on gene expression (60,61). Cooperative interactions involving C/EBP␤ are important for the ability of thyroid hormone, glucocorticoids, and cAMP to stimulate the PEPCK promoter (8,10), and a complex containing C/EBP␤ can interact with IRSs which function cooperatively to enhance effects of glucocorticoids on activity in the IGFBP-1 and PEPCK promoters (11,62,63). C/EBP proteins appear to play an important role in mediating effects of insulin on stimulated PEPCK promoter activity that are independent of an IRS (9), and overexpressing a fragment of p300/CBP that contains the C/EBP␤-binding domain disrupts the ability of insulin to suppress PEPCK gene expression in cAMP-stimulated hepatoma cells (23). Based on the present study, it is interesting to speculate that signaling to this region of p300/ CBP may contribute to the ability of insulin or related growth factors to suppress promoter activity by disrupting cooperative interactions involving C/EBP␤.
In this context, it is important to note that this region of p300 (1752-1859) also interacts with other factors that are known to be important in the regulation of gene expression. Nasrin et al. (31) recently reported that DAF-16, a forkhead protein in Caenorhabditis elegans, interacts with this region and the KIX (CREB binding) domain of p300/CBP (31), and pull-down studies indicate that FKHR, a mammalian homologue of DAF-16, also interacts with these regions of p300/CBP. 3 It will be important to determine whether signaling to p300/CBP may contribute to the ability of insulin to suppress the function of the C-terminal transactivation domain of these forkhead proteins (64).
As recently reviewed, several lines of evidence suggest that 3 S. Guo and T. Unterman, unpublished observations. p300/CBP proteins play an important role in the regulation of the cell cycle and cell survival (65). C/EBP proteins have important effects on the proliferation of hepatocytes (55), and the region of p300/CBP that binds C/EBP␤ (17) also interacts with other factors that are involved in the regulation of cellular growth and differentiation, including E1A (18), cyclin E-Cdk2 kinase (22), pp90 Rsk (23), SV40 large T antigen (24), c-Jun (25), c-Fos (26), MyoD (27), YY1 (28), and Ets-1 (29). Based on these observations and the results of the present study, it will be important to examine the role that signaling through PI 3-kinase and PKB to this region of p300/CBP plays in mediating effects of insulin and related growth factors on cellular proliferation and differentiation, and metabolism.